Spend five minutes on YouTube and you will encounter the same narrative on a loop: testosterone is in freefall, men are in hormonal crisis, the numbers are catastrophic. It is compelling content. It also requires a harder look at what the data actually says.
A 2026 analysis of four National Health and Nutrition Examination Survey (NHANES) cycles — covering 10,357 U.S. men over the years 2011 to 2023 — found something that runs counter to the prevailing alarm. The age-standardized prevalence of testosterone deficiency (defined as total testosterone below 300 ng/dL) declined meaningfully over that period. Not a small noise-level shift. A real drop.
That finding deserves careful reading. Because the data tells a nuanced story, and the nuance matters if you are trying to understand your own hormonal health rather than just consume content about it.
What the NHANES Analysis Actually Found
The study, published in the American Journal of Men's Health by Goulian, Ashkezari, and Eleswarapu, analyzed men aged 18 and older across NHANES cycles from 2011 through 2016 and again from 2021 through 2023. The primary outcome was the age-standardized prevalence of testosterone deficiency, using the widely cited threshold of total testosterone below 300 ng/dL.
The headline finding: deficiency prevalence dropped from 29.3% in the 2011–12 cycle to 22.4% in 2021–23. That is a roughly 7 percentage-point decline in about a decade.
This is not measuring whether testosterone levels in the general population are rising. It is measuring what fraction of men fall below a clinical threshold — the cutoff below which a physician would typically consider a diagnosis of hypogonadism. Those are different things, and conflating them is how most of the discourse goes wrong.
The NHANES cycles used: 2011–12, 2013–14, 2015–16, and 2021–23. Total sample: 10,357 men ≥18 years. Testosterone measurements used serum total testosterone, with deficiency defined as below 300 ng/dL — a standard clinical threshold. DOI: 10.1177/15579883261433808.
The Distinction That Almost Nobody Makes
Here is where most commentary goes off the rails. There are two separate questions:
- Has the clinical deficiency prevalence changed? (What fraction of men fall below 300 ng/dL?)
- Have average testosterone levels in the population changed over time? (Where is the distribution centered?)
The NHANES analysis addresses question one. The answer, in recent data, is: deficiency prevalence has declined.
A separate and older body of literature — the Travison et al. study published in the Journal of Clinical Endocrinology and Metabolism in 2007 is the landmark reference — addresses question two. That research found that average serum testosterone levels in American men declined substantially between the 1980s and the early 2000s, by roughly 1% per year, in ways that could not be explained by aging alone. The secular decline documented there was real and has been replicated in other cohorts.
Both findings can be true simultaneously. Mean testosterone levels may be lower today than they were in the 1980s — the distribution shifted left — while at the same time, recent trends show fewer men falling below the clinical deficiency threshold. How? The most likely explanation involves the behavioral risk factors that predict deficiency, and what happened to them.
When you see claims about "testosterone levels declining," check whether the source is discussing population mean levels across decades, or current prevalence of clinical deficiency. They are different metrics from different studies. The first has strong evidence. The second, per this 2026 data, is moving in the opposite direction.
What Drove the Decline in Deficiency?
The NHANES data does not just report the trend — it identifies behavioral and metabolic predictors. Two factors stand out.
Current smoking declined substantially. Smoking is one of the more reliable suppressors of testosterone production, operating through oxidative stress, Leydig cell function, and SHBG dynamics. Population smoking rates in the U.S. dropped significantly over the period analyzed. The study found current smoking was a meaningful predictor of deficiency status.
Sedentary time decreased. The fraction of men reporting 8 or more hours per day of sedentary time declined over the study period. Physical inactivity is robustly associated with testosterone deficiency through multiple mechanisms: adiposity, insulin resistance, systemic inflammation, and the direct anabolic stimulus of movement. Less sedentary time means fewer men in the metabolic conditions that drive T below clinical thresholds.
Metabolic health more broadly — weight, insulin sensitivity, cardiovascular risk factors — is the context in which testosterone operates. When population-level metabolic health improves on specific dimensions, deficiency prevalence can fall even if the absolute distribution of T levels is not dramatically different from what it was a decade ago.
What This Means for Your Modifiable Variables
The behavioral predictors identified in this data map directly onto what the American Heart Association calls Life's Essential 8: diet, physical activity, sleep, nicotine exposure, body weight, blood glucose, blood lipids, and blood pressure.
These are not abstract public health recommendations. They are the variables that show up repeatedly in studies on testosterone deficiency. Smoking is in there. Sedentary time is in there. Adiposity and metabolic health are in there. Sleep quality is heavily implicated in testosterone production — testosterone release is predominantly nocturnal and tied to slow-wave sleep architecture.
The practical implication: a substantial portion of clinical testosterone deficiency in men is not fate. It is a downstream consequence of modifiable behaviors. The men in this NHANES data who fell below 300 ng/dL were, in aggregate, smoking more, moving less, and carrying more metabolic burden.
The four behaviors with the strongest evidence for supporting healthy testosterone levels are: not smoking, maintaining a healthy body composition, getting consistent and sufficient sleep, and regular resistance or aerobic exercise. These show up in the testosterone literature repeatedly, and they are all within your control to track.
For a deeper look at specific behavioral mechanisms, see our posts on how sleep shapes testosterone production and the role of endocrine disruptors.
Running Your Own N-of-1
Here is where population-level data has limits and where individual tracking becomes relevant. The NHANES finding tells you about trends across tens of thousands of men. It does not tell you where your testosterone is, whether it has changed, or what is driving it in your specific case.
That requires your own data.
A practical tracking protocol:
Quarterly lab work. Total testosterone and free testosterone, drawn in the morning between 7 and 10 AM, fasted, after a normal night of sleep. This timing matters — testosterone follows a strong diurnal rhythm, with peak levels in the early morning. An afternoon draw can easily look 15–20% lower than a morning draw in the same person.
Behavioral variables that map to the NHANES predictors. In parallel with lab work, track the inputs: sleep duration and quality (a wearable gives you objective data, or you can log subjective scores daily), daily step count or exercise sessions as a proxy for sedentary time, smoking status, body weight, and any relevant metabolic markers if you get a full panel.
Correlate over time. This is the step almost nobody does, because without a structured tracking system, the data sits in scattered lab PDFs and forgotten notes. The value is in lining up your testosterone numbers against your behavioral inputs from the preceding weeks and looking for patterns. Did your T drop in the quarter when your sleep was terrible? Did it hold steady when you were exercising consistently?
Track your Life's Essential 8 behaviors for 90 days alongside a quarterly lab draw. The goal is not to diagnose anything — it is to understand your own data well enough to know whether behavior changes are moving your numbers.
This is exactly the kind of structured self-experiment that Prova is built for. You define the variables, log daily data, get quarterly lab anchors, and let the pattern emerge over time. That is a more useful signal than any single testosterone reading in isolation.
Be the first to try Prova
We're building an app to track whether testosterone-tracking actually works. Join the waitlist.
Connecting This to the Broader Picture
The NHANES finding does not mean the testosterone crisis content is entirely wrong. Population mean testosterone levels being lower today than they were fifty years ago — if you accept the Travison literature — is a real concern. The secular environmental and behavioral pressures that drove that shift (obesity, sedentary lifestyles, endocrine-disrupting chemical exposure, poor sleep) have not disappeared.
What the 2026 data suggests is that the clinical deficiency end of the spectrum has improved recently, likely driven by declines in smoking and sedentary time. That is genuinely good news. It also clarifies the mechanism: the biggest drivers of testosterone deficiency are behavioral and metabolic, not simply age or some irreversible hormonal doom.
The men reading this in their 30s and 40s are not helpless passengers in a hormonal decline. The data suggests that what you do with your sleep, your movement, your metabolic health, and your smoking status has a measurable effect on whether you end up in the clinical deficiency category.
Understanding early signs of low testosterone is useful context here — but the more important question is whether the inputs you can control are optimized. For a full overview of what the evidence says about nutritional support, see our post on testosterone support supplements, which covers what has actual evidence versus what is marketing.
Bottom Line
The narrative that testosterone deficiency is an exploding crisis across American men is not well-supported by the most recent NHANES data. Clinical deficiency prevalence declined from 29.3% to 22.4% between 2011 and 2023. Smoking and sedentary time were the behavioral predictors most associated with that shift.
This is distinct from the question of whether mean population testosterone levels have declined over decades. That finding, from an older literature, remains real. The two metrics measure different things and can move in different directions simultaneously.
What the data collectively supports: testosterone is highly sensitive to behavioral inputs. Smoking, sleep, movement, body composition, and metabolic health are not peripheral lifestyle factors — they are the primary drivers of where your testosterone lands. That is worth tracking. And it is worth tracking against your actual lab numbers over time, not just against population averages that may or may not apply to you.